We have continued studies of chromatin structure and the regulation of eukaryotic gene expression by analysis of ATP-dependent chromatin remodeling enzymes. We completed studies on the biology of ATP-dependent chromatin remodeling by NURF (Nucleosome Remodeling Factor) in the mouse. Bptf, the largest subunit of NURF, is required for maturation of thymocytes. We traced this defect to the misregulation of a number of genes important for thymocyte development, and found Bptf-dependent changes in chromatin structure at gene targets as well as a requirement for NURF in the binding of the sequence-specific transcription factor SRF to chromatin. This work was submitted, revised, accepted for publication, and is now published in Genes and Development. We have also completed a study on the mechanism of histone H2A.Z replacement by the SWR1 chromatin remodeling complex. We found that promoter-proximal nucleosomes are highly heterogeneous for H2A.Z in Saccharomyces cerevisiae, with substantial representation of nucleosomes containing one, two, or no H2A.Z molecules. SWR1-catalyzed H2A.Z replacement in vitro occurs in a stepwise and unidirectional fashion, one H2A.Z-H2B dimer at a time, producing heterotypic nucleosomes as intermediates and homotypic H2A.Z nucleosomes as end products. The ATPase activity of SWR1 is specifically stimulated by H2A-containing nucleosomes and free H2A.Z-H2B dimers, leading to eviction of nucleosomal H2A-H2B and deposition of H2A.Z-H2B. These results suggest that the in vivo substrates govern the specificity and outcome of the replacement reaction. This work was submitted, revised, accepted for publication, and is now published in Cell. We have continued investigations on the molecular architecture of centromere-specific nucleosomes containing histone variant CenH3. We found that DNA base composition influences reconstitution of nucleosomes containing Saccharomyces cerevisiae CenH3 (Cse4). Reconstitution is robust for non-centromere DNA, but inefficient for centromere DNA, due to an AT-rich centromere element that hinders association of Cse4-containing histone octamers. By contrast, inclusion of nonhistone Scm3 with Cse4/H4 permits reconstitution on natural centromere sequences. Scm3 has a DNA binding domain that exhibits preference for the AT-rich centromere element, and a histone chaperone-like domain that promotes loading of Cse4/H4. In vivo, Scm3-GFP localizes to the centromere cluster in all phases of the cell cycle, including mitosis. Scm3 can also be cross-linked to centromeres in synchronized cells throughout the cell cycle, even when Cse4/H4 is temporarily dislodged in early S phase. These findings suggest a model in which centromere DNA-bound Scm3 aids recruitment of Cse4/H4 to assemble an H2A/H2B-deficient centromeric nucleosome. This work was submitted, revised, accepted for publication, and is now published in Molecular Cell.